Plugins

One-click ✨ polish button for the composer: rewrites prompts and drafts with WorkBuddy humanizer-style rules (kills AI-flavored phrasing, varies rhythm, mode-aware for instructions/questions/prose) via free Zhipu GLM-4.5-Flash or any OpenAI-compatible API, falling back to offline local rules.

Lets sandboxed agents reach the web through your local proxy: net_fetch fetches pages, net_proxy_status reports proxy status.

Renders the built-in cache-hit percentage with three decimals and widens the composer stats line to prevent truncation.

Enhanced auto-compaction defaults for DeepSeek Harness agent presets: apply tuned threshold/retention budgets and per-model policies with one command, idempotent with backups and rollback.

Conversation history recall for DSH: three-layer (literal/fuzzy/semantic) retrieval over every past session's original text, fully local & offline — AI never forgets what you told it. One-command install via `dsh.bundle.patch`, semantic layer runs in a worker thread.

QQ 2007-inspired DSH Web skin: 72 native theme tokens, scoped three-pane chrome, original offline art and pixel buddy, an optional synthesized send chime, responsive/accessibility fallbacks, and reversible Settings switches.

Common Word Box (常用词盒子) — DeepSeek Harness web client plugin: a persistent slide-up word/phrase panel next to the composer input, with global/current-project buckets and one-click insert. Works in every workspace.

Upload local images and files to your image host through PicGo's existing configuration (PicGo Cloud, GitHub, S3, COS, Qiniu, or any installed uploader plugin), via a `picgo_upload` tool and a `/picgo` command.

Native LLM-provider vision bridge: images pasted in the chat are described by a vision model (Qwen3-VL via pi-ai/llama.cpp) and the text description is fed to text-only DeepSeek for the reply — image admission, routing and compaction all run through harness-native mechanisms, with an LRU description cache and 503 retry.

Generate images through a logged-in web ChatGPT session and download them to a local directory.

Native DeepSeek Harness browser-agent plugin providing session-isolated Chromium automation with CDP and DOM tools.

Local multi-plan bridge for DeepSeek Harness: use your existing ChatGPT Codex, Claude, ZCode (GLM) and Qwen subscriptions through one loopback proxy, with a settings-page status card and one-click provisioning. No dsh core changes.

Fork companion for DSH: on any official fork, copies every direct subagent child of the source session onto the forked conversation (recursively, whole subtrees) through the official agent API, leaving each copy as a cold session the official resume path can wake — so the new conversation keeps its subagent panel and can keep messaging those children, while the originals stay with the old branch.

Configurable send and newline keys, a composer context menu with three sources (leave it to DSH and other plugins / browser menu / plugin-styled custom menu), a quick-command panel with categorized prompt lists where each prompt can append on every send or only on the first message of a session, a create-and-move row at the bottom of each category, a prompt optimizer that rewrites the draft with a separate model call, settings-panel size memory, and automatic x-opencode-session header injection for OpenCode routes. Prompt lists live in a global ~/.dsh/quick-prompts.json that is independent of the session and the project.

Runs PowerShell scripts through base64 encoding, so quotes, $, backslashes and CJK survive as written. Same sandbox channel as the built-in pwsh tool.

Unit conversion toolbox covering 20 categories — length, mass, temperature, area, volume (incl. US cooking units), speed, time, data sizes (decimal MB vs binary MiB), data transfer rates (Mbps vs MB/s), acceleration incl. g-force, illumination (lux / foot-candle), pressure, energy, angle, frequency, power incl. three horsepower definitions, force, torque, typography (px/pt/em/rem), and fuel economy (mpg ↔ L/100km). Two tools — convert_unit and list_units — zero runtime dependencies, pure arithmetic; each result includes the formula applied.

LLM provider adapter that routes model calls through any ACP (Agent Client Protocol) server — Claude Code, Codex, Gemini CLI, Devin and more — with a registry browser, per-server settings UI, lazy authentication, and dynamic model discovery.

Configurable image recognition for text-only DSH models: image messages are first transcribed by an OpenAI-compatible vision model (Base URL, model ID and API key set in a Settings section) and then passed to the main model as text, while image-input support is advertised. The API key auth scheme is selectable — OpenAI, Anthropic, Gemini or Azure style request headers — and a missing key is reported before any request is sent.

Routes the DSH web_search tool through Ollama's web search API, reusing OLLAMA_API_KEY. No DeepSeek official key needed.

Drag-and-drop file upload into the conversation composer, token usage pill with quota tracking, and session archive management — a three-in-one daily-driver addon.

DSH Desktop enhancement plugin: task-execution discipline plus a real relationship, per session. Discipline: adaptive protocol tiers (short for chat / full for tasks / lean for reasoning models), intent routing across Q&A, lookup, discussion, diagnosis and execution (diagnosis never fixes unasked), evidence recency (verify timestamps and causality before citing logs or old errors), real tool-result verification (failed or unknown results are never reported as done), self-healing on repeated failures with a cross-session reflection loop, compaction awareness that re-anchors state after the host condenses history, and document-capability awareness (probe for document tools and constrain accordingly: with tools, read before you write and read back before claiming done; without them, state the boundary instead of hand-parsing binary office files). Persona: distill named characters from chat logs, novels, scripts and character-setting documents, with tone, catchphrases and reply length anchored to real material statistics; long-term memory with automatic expiry for time-bound facts, corrections captured as style rules, approved replies pinned as corpus, two-way feedback that converges with use, memory star map, and persona card export/import. Constraints are injected only when relevant, so idle chat pays no extra tokens.

Agent discipline guard for DeepSeek Harness: four gates enforced at the harness boundary — loop circuit-breaker, token-cost fuse, route-switch warning, plan approval — where prompt-level reminders can't reach, so an agent can no longer burn a weekly quota in minutes; one-click 🛡 chip, hot-reloadable thresholds, and presets from invisible-but-armored to warn-only.

Three-pane workspace layout with file-tree browsing, file editing, a non-linear conversation mindmap view, and mobile mode support.

Kimi WebBridge host plugin for DeepSeek Harness: model tools that drive the user's REAL browser (navigate, snapshot, click, fill, evaluate, screenshot, network inspection, tab management) through the local kimi-webbridge daemon